A fresh Take on ‘Barker Dynamics’ for MCMC

DOI10.1007/978-3-030-98319-2_8MaRDI QIDQ6154290zbMATH OpenOpenAlexFDO

Authors Samuel Livingstone, Giacomo Zanella

Publication date 14 February 2024

Published in Springer Proceedings in Mathematics & Statistics (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2012.09731

zbMATH Keywords

Metropolis-HastingsMCMCMarkov chainsLangevin dynamicsBarker dynamics

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Monte Carlo methods (65C05)

Abstract: We study a recently introduced gradient-based Markov chain Monte Carlo method based on 'Barker dynamics'. We provide a full derivation of the method from first principles, placing it within a wider class of continuous-time Markov jump processes. We then evaluate the Barker approach numerically on a challenging ill-conditioned logistic regression example with imbalanced data, showing in particular that the algorithm is remarkably robust to irregularity (in this case a high degree of skew) in the target distribution.

Recommendations

• Barker's algorithm for Bayesian inference with intractable likelihoods
• The Bouncy Particle Sampler: A Non-Reversible Rejection-Free Markov Chain Monte Carlo Method
• Optimal scaling of MCMC beyond Metropolis
• Optimal search efficiency of Barker's algorithm with an exponential fitness function
• MCMC using Hamiltonian dynamics

Cites work

• A note on Metropolis-Hastings kernels for general state spaces
• Equation of state calculations by fast computing machines
• Exponential convergence of Langevin distributions and their discrete approximations
• General state space Markov chains and MCMC algorithms
• Information-geometric Markov chain Monte Carlo methods using diffusions
• Informed proposals for local MCMC in discrete spaces
• Kinetic energy choice in Hamiltonian/hybrid Monte Carlo
• Leave Pima Indians alone: binary regression as a benchmark for Bayesian computation
• MCMC for imbalanced categorical data
• Monte Carlo sampling methods using Markov chains and their applications
• Nonasymptotic convergence analysis for the unadjusted Langevin algorithm
• Optimal Scaling of Discrete Approximations to Langevin Diffusions
• Optimum Monte-Carlo sampling using Markov chains
• Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors' reply
• Some properties of skew-symmetric distributions
• The tamed unadjusted Langevin algorithm
• User-friendly guarantees for the Langevin Monte Carlo with inaccurate gradient

This page was built for publication: A fresh Take on ‘Barker Dynamics’ for MCMC

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6154290)